Comprehensive Summary
Rakaee et al focuses on how machine-learning estimates of tumor-infiltrating lymphocytes (TILs) from hematoxylin-eosin images compare to immune checkpoint inhibitors in patients with non-small cell lung cancer (NSCLC). Artificial intelligence models are particularly valuable in analyzing spatial patterns and can help quantify TILs on routine hematoxylin-eosin stained slides for NSCLC. A total of 685 patients with advanced-stage NSCLC at Dana-Farber Cancer Institute and Imperial College London were treated with anti-programmed death ligand 1 (PD(L)1) monotherapy between February 2014 and September 2021. Image analysis was performed on whole slide hematoxylin-eosin images of the biopsy and samples using supervised machine-learning algorithms. Data was collected on TIL cell count, tumor mutational burden (TMB), PD-L1 expression, and clinical response to ICl therapy. Overall, there was a better survival and response outcome of patients with higher TIL levels receiving ICl monotherapy. Using weighted linear regression and comparing the biomarkers, TIL levels showed stronger and independent association with clinical benefit and better response to ICl therapy than TMB levels. TIL levels were successfully quantified through a machine learning approach with the stained slides.
Outcomes and Implications
Information about predictive biomarkers to ICl therapy is limited, so this research on patient TIL assessment can help identify biomarkers such as TIL and TMB levels that can guide precision therapy. Patient TIL assessment using machine learning in this study is very cost-effective and can easily be scaled for routine clinical care. Using artificial intelligence also has minimal user intervention and reduces manual scoring and pathologist bias. This can be integrated into pathology laboratories and help speed up pathology reports on cell and tissue biopsies. TIL can be used as a biomarker to track in the clinic to see how effective ICl monotherapy could be in a patient with NSCLC. If a patient has higher TIL counts and has a PD-L1 negative advanced stage NSCLC, ICL therapy could be a good choice to improve overall survival rates. More research needs to be done on machine learning based TIL profiling in patients with earlier stage NSCLC.